Comparing prehistoric constructed languages: world-building and its role in understanding prehistoric languages

Abstract

In this paper, we compare the languages each of the authors invented as prehistoric languages for popular culture media. Schreyer's language, Beama, was created for the film Alpha (2018), while Adger's language, Tan!aa Kawawa ki, was created for a television series on how early hominins spread throughout the world (the series was green-lit but then cancelled). We argue that though this creative process may seem far removed from classical research paradigms on language evolution, it can provide some insight into how disparate research on the possible properties of prehistoric languages can be brought together to illustrate how these languages might have worked as whole linguistic systems within these imagined worlds, as well as in prehistory.

This article is part of the theme issue ‘Reconstructing prehistoric languages’.

1. Introduction

The question of how humans developed natural languages and what such languages may have been like is fraught with well-known difficulties. However, these are questions that are easily answerable for constructed languages. Constructed languages, also known as conlangs, are languages that an individual, or a group of individuals, develop for a specific purpose.¹ Contemporary conlangs are usually developed for one of three purposes: (i) to be an international auxiliary language, known as an auxlang; (ii) to be a language used for an artistic endeavour, known as an artlang; or (iii) as a means of exploring certain philosophical concepts through constructing a language that embeds them, sometimes known as an engelang [2]. Constructed languages have also been categorized in terms of their method of creation. An a posteriori language is one that has been developed using aspects of other naturally occurring languages, while an a priori language is one that is created without intentional use of specific languages. While constructed languages have, in the past, been considered ‘fake’, and hence uninteresting, academic study of these languages has developed in the last decade with the publication of a few key works in this area, focusing on the languages' communities of users and how these languages can be used for educational purposes (e.g. [1,3,4], and more recently [5]). Research on constructed languages has also looked at learners of constructed languages that are associated with literature, TV and film, including Weiner & Marshall [6] on Tolkein's Elvish conlangs, Wahlgren [7] and Okrand et al. [8] on the conlang Klingon, Schreyer [9] on Na'vi from the film Avatar, and Peacey [10] on Trigedasleng from the television show The 100, complementing earlier research on learners of Esperanto [11–13].

In this paper, we take a different tack, and argue that conlanging (the practice of developing conlangs) can shed some light on what prehistoric languages may have been like, using our own experiences developing two prehistoric artlangs developed for media as examples. Schreyer's language, Beama, was created for the film Alpha (2018), while Adger's language, Tan!aa Kawawa ki, was created for a television series on how early hominins spread throughout the world (the series was green-lit but then cancelled). Through these examples, we show how bringing together disparate kinds of research can provide a picture of how prehistoric languages may have worked.

In the introduction to his book From Elvish to Klingon, Adams [3] describes what he calls ‘the spectrum of invention’. This spectrum outlines the many reasons why individuals have taken it upon themselves to develop new languages. He outlines the Western world's early preoccupation with finding the world's original language, writing:

Ever since [the fall of Babel], language engineers have attempted to build not just a better language, but the original language, in order to unify the peoples of the world, and let's be honest, to regain Eden and our innocence, to reach a perfect human state by means of unambiguous speech and clear communication. [3, p. 4]

In contrast with this, we were tasked with contributing to the world-building of the media productions we were involved with. World-building is the practice of developing an alternative reality for a piece of literature, television or film. The World Build Institute describes world-building in the following way:

World Building is founded on three beliefs, namely that storytelling is the most powerful system for the advancement of human capability due to its ability to allow the human imagination to precede the realization of thought; that all stories emerge logically and intuitively from the worlds that create them; and that new technologies powerfully enable us to sculpt the imagination into existence. [14]

The productions we were involved with were keen to provide an authentic experience for viewers of the world of early humans and thereby ‘sculpt the imagination into existence’. For the aspects of the productions that were connected to language, this meant developing a whole novel linguistic system, in the context of current knowledge from theoretical and descriptive linguistics about how languages work. By doing this, it is possible to develop a cohesive system, and hence to infer a potentially authentic example of a prehistoric language, by exploring how choices made about individual parts of the language interact with each other in the context of knowledge from linguistic theory, linguistic typology, historical linguistics, linguistic anthropology, language acquisition and the cognitive science of language. In other words, the production teams in these instances were not focused on whether the language sounded ‘exotic’ or ‘alien’, which was true in other cases where languages have been made for popular media about non-human beings, such as Klingon for Star Trek [8], but whether or not the languages were authentically human.

An analogy from how we develop other prehistoric knowledge may be useful in seeing how this might work, keeping in mind that human language is both biological and social in nature. On the biological side, a palaeontologist might look at a fragment of an ancient fish bone and combine it with knowledge of fish anatomy, development and genetics to infer what the whole fish might have been like, and from there hypothesize about the behavioural ecology of that organism. On the social side, an archaeologist might investigate the structure of a Palaeolithic hand-axe and infer what kinds of technology and processes might have been used in its manufacture, and then more speculatively, what its uses might have been, using the minimal data point the hand-axe gives in combination with wider theoretical context from anthropology plus knowledge of what people do in the present day (for example, information from modern flint knappers). Applying these analogies, building a conlang around plausible features of a prehistoric language is a way to develop a holistic understanding of what a whole linguistic system might have been like, inferring from how those plausible features interact with information we have about how human languages are structured and how they are used.

In the next sections, we will describe the processes of developing our prehistoric languages and the motivations for the choices we made based on academic and artistic restrictions, connecting these choices to research on prehistoric languages. An interesting side note to our arguments about the role of conlanging as a method of inquiry into language evolution is that languages developed for pop culture often garner media attention [15,16], which provides a window for non-linguist audiences to become aware of the intellectual issues surrounding language evolution and linguistics more generally [17].

2. The construction of Beama

In 2015, I (Schreyer) was contacted by Andrew Rona, a producer for the film that would become Alpha (2018), to see if I was interested in developing a new language for a film set in prehistoric Europe, 20 000 years ago. While I had developed another conlang for media, Kryptonian for the film Man of Steel (2013), this language was spoken by aliens and I had never attempted to make a language that was intended to represent a prehistoric language that would have been spoken by humans. As an anthropologist, trained in all four fields of anthropology (linguistic, cultural, biological and archaeological), I was particularly intrigued about using my overall knowledge from anthropology to develop this early prehistoric language. As this language, which eventually became known as Beama (pronounced as [beama] in the International Phonetic Alphabet (IPA) and translated as ‘our talk’), was intended to be a once spoken natural language, I developed this conlang as an a posteriori language using information from proto-languages. I also classify this language as an artlang since it was made with the intention to be included in a film.

The characters in the film that spoke Beama were described by the production team as Cro-Magnons or anatomically modern Homo sapiens and, while most scholars agree that Cro-Magnons had language in the Upper Palaeolithic, few have suggested what this might be like as it is difficult to predict how languages may have changed over such lengthy periods of time. However, scholars have developed three proto-languages from close to this time period and geographical area and, in order to make this constructed language as linguistically plausible as possible, and authentic for the viewer, I considered these in order to develop the phonology (sound system), morphology (word structure) and syntax (sentence structure) of Beama. These proto-languages were: Proto-Nostratic (dated to around 15 000–12 000 BCE, which is sometimes described as ‘Epipalaeolithic’ or what is spoken on the way to the Mesolithic era) [18], Proto-Eurasiatic (dated to around 15 000 BCE as well, but an alternative grouping to Nostratic) and, finally, Proto-Dené-Caucasian, dated to 20 000–25 000 BCE, pre-Eurasiatic migrations found in refuge areas at the last Glacial Maximum) [19].

(a). Phonology

All three of these proto-languages had remarkably similar sounds and I ended up using a blend of what is considered to be in each (see tables 1 and 2). However, I also simplified some of the sounds so that the actors, who were for the most part speakers of English, could pronounce them with ease. One thing to note here is that I included three ejective consonants, as I knew that the production team had imagined the landscape of the world of Alpha as having both plateaus and mountains. Therefore, I considered [20] work on the geographical influences of high altitudes and retained these sounds in the language's phonology as well.

Table 1.

Beama consonants.

	bilabial	alveolar central	alveolar lateral	post-alveolar	palatal	velar	glottals
stops
ejectives	p’	t’				k’
voiceless	p	t				k	ʔ
voiced	b	d				g
nasals	m	n				ŋ (‘ng’)
fricatives
voiceless	f	s	ɬ	ʃ (‘sh’)		x	h
voiced	v	z		ʒ		ɣ	ɦ
affricates
voiceless		ʦ		ʧ (‘ch’)
voiced		ʣ		ʤ (‘j’)
central	w				j (‘y’)
approximants		ɹ
lateral approximants		l

Table 2.

Beama vowels.

	front	centre	back
high	i		u
mid	e		o
low	æ	a

In order to train the actors, I provided charts of the sounds written in the IPA, as well as their linguistic description. I trained the actors to read the IPA characters and also provided them with audio recordings. Some of my decisions on the language were due to conversations I had with Albert Hughes, the director of the film. For instance, during an early discussion, Albert mentioned that two languages that are pleasant-sounding to his ears are Spanish and Italian. These languages tend to have shorter syllables with more vowel sounds. Information from the proto-language of Dené-Caucasian states that root words often have the syllable structure of CV–VC–V, CV–VC–VC–V, VC–V, VC–VC (where C is a Consonant and V is a Vowel) [19]. This formation made for melodic-sounding root words and I chose this as the syllable pattern. Dené-Caucasian also places stress on the penultimate syllable of root words and this was the pattern for stress I included as well. This was marked in the written transcripts for the actors so they were able to easily see and then pronounce the stressed syllable. It is very important that there is consistency on stress between actors for the language to sound realistic to an audience's ears.

(b). Morphology and syntax

For the word structure of Beama, I relied heavily on Bomhard's 2014 volume A comprehensive introduction to Nostratic comparative linguistics: with special reference to Indo-European. In particular, after developing my syllable structures above, I used the English–Nostratic index at the back of this book, in order to see what a particular word was in Proto-Nostratic according to Bomhard, and then modified according to the principles I had set up for the system. However, many times the words I needed were not present in the index and so these I developed based on similar patterns. To determine the order of words in sentences, I also examined the information on the proto-languages and chose the sentence order of subject-object-verb or subject–verb (when no object is present) as this is the suggested sentence order of Proto-Nostratic [21].

The film Alpha was entirely in the language of Beama (and subtitled for audiences). As a result, the production team requested lines that were often grammatically complex. To translate these lines, I developed a complex pronominal system (table 3), which included both inclusive and exclusive first person plural and animate/inanimate divisions for third person singular.

Table 3.

Beama pronominal system.

1 singular	mi	1 plural (inclusive)	maa
		1 plural (exclusive)	na
2 singular	ti	2 plural	ta
3 singular (animate)	zi	3 plural	za
3 singular (inanimate)	ze

^aCuriously, Adger's first person plural form is the same (see §3 on his language).

Some of the lines requested for the actors also had instances of embedded clauses. For example, the sentence ‘Now I know why you came with me’ is translated in the following way:

(1)	[ki	mi.na	mu.a.ko.ti	e.ya.mi	nu]2
	why	1st.sing.loc.	to come.past. 2nd. sing	to know. 1st. sing.	now

I used framing particles ki and nu, but the verbal subject markers (second singular -ti and first singular -mi), as well as tense (ko—past tense), are suffixes, as is the locative marker na. Conditional clauses worked in a similar way; the sentence ‘If the great beast walks through fire, we walk through fire’ translates as:

(2)	[ɬe	duɹa	biɹa	paha.na miɹa.sa	paha.na miɹa.ma]
	if	beast	great	fire.loc. to walk. 3rdpl ani.	fire. loc. to walk. 1st. pl.

Again, a framing particle to indicate conditionality is used at the beginning of the phrase, łe, and then the phrases are parallel. The great beast is the Beama speaker's word for bison, which they hunt every year and which is a major plot point to the movie.

In order to help give a sense of authenticity in this created world, speakers of the Beama language also had expressions of surprise or what could be considered swear words or (expletive) speech automatisms (see [22]). In particular, the language includes: aargh (uuuaaas), oh my! (uki), hey (æt'a) and ummm (iyeee). In these examples, vowels are often held longer than the regular CV patterns, but this is also consistent with similar words in natural languages. It is rare that such words would be created for an artlang, but as the film Alpha was entirely in the language of Beama, lines were developed for non-main characters in all scenes with more than one actor. Many of these actors in the background were involved in daily tasks from the world of Alpha, such as sitting around a campfire when an animal attacks, and hence these words were born. They also contribute to the authenticity of this world, as does the inclusion of metaphor and metonymy. Lakoff & Johnson in their seminal volume Metaphors we live by write, ‘… human thought processes are largely metaphorical …metaphors as linguistic expressions are possible precisely because there are metaphors in a person's conceptual system’ [23, p. 6]. For the Beama language, when an individual has died, they are imagined to travel to another world, which is called ‘the other-side forest’ or [imaɹa.na tina] (literally, forest-locative other).³ Speakers of Beama live their lives in forests, which is the location of their main camps, and so imagining life after death as another forest is an apt metaphor. For anthropologists, ‘the study of mortuary practices and rituals provides [them] with rich and varied insights into a community's conceptualization of mortality, social memory, and relationships between individuals’ [25, p. 7], and including a sense of how this early imagined human culture viewed death is a part of the world-building of the film. In the next section, we discuss how Adger developed his prehistoric language before discussing the similarities and differences between the processes each of us employed, as well as the resulting outcomes.

3. The construction of Tan!aa Kawawa ki

I (Adger) was approached in early 2017 by a production company that was developing a drama documentary which would involve encounters between early Homo sapiens and other hominins (Homo neanderthalensis, Homo floresiensis, Homo denisova and Homo naledi). The company had approached Maggie Tallerman as an expert on language evolution, and Tallerman contacted me, as she knew I had experience of developing conlangs for TV programmes (I had developed two conlangs for the ITV series Beowulf: Return to the Shieldlands, and document some of this in [17]). It became clear that the production team wanted upwards of 10 conlangs, and so I coopted my colleague Coppe van Urk, with whom I had been developing an educational programme for schools using conlangs [26].

The production company was keen to have the conlangs developed on the basis of the latest research on what early human languages may have been like. This meant that the conlangs would be a priori: not based on particular real or reconstructed (hypothesized) languages, but rather based on research about early human language. Like Beama, the languages would be artlangs, with input from the production team, and their purpose was to add authenticity to the world of early humans as depicted in the series and as informed by archaeological and palaeoanthropological research. The time span that the drama documentary was to cover was all within the last 75 000 years, and our team followed Hurford [27] in taking human language in its basic form to have emerged between 100 000 and 60 000 years ago. This meant that the Homo sapiens languages would be structurally similar to current human languages, though perhaps with some hallmarks of being young languages. We also developed a sketch of design choices for the languages of Neandertals, Denisovans and Homo floresiensis, but the series was cancelled in late 2018 before these were developed in any depth.

I was given an early script of an episode and used this in combination with the various design choices that the conlang team had agreed on, plus further decisions about specifics that I made myself, to develop a grammar for the language spoken by the hominins in that episode. Some of the general choices were made by the conlang team on the basis of properties of the human articulatory system, some on the basis of the geographical spread of linguistic properties, and commonly adopted hypotheses about where humans originated, and some on the idea that human cognition has certain defaults for language, and that these defaults were likely to be evident in early human language, before being overwritten, in some languages, by the effects of time, contact with other languages, and social and cultural change [28].

Because we had decided that these languages were relatively young, one overarching choice was to develop as minimal a system as possible that would still allow modern-day levels of expressive capability. This meant minimal phonological inventories, simple syllable structures and a simple but expressive syntax and system of word meanings.

(a). Phonology

In terms of sound systems, the team decided to use a highly nasalized set of sounds. Deaf babies, when they babble orally, use more nasal sounds than non-deaf babies do [29]. This is plausibly because they have no evidence for the preponderance of oral sounds in their input, and their anatomical and cognitive states make nasality a kind of default. Further, the usual airflow in normal breathing is a nasal airflow. We made a design choice that nasality would be a prominent feature of the sound system.

We also adopted the idea that the sound system would have at least some clicks. It is plausible that there was a genetic split between San Africans and other humans well over 100 000 years ago, and even today, the languages spoken by San Africans (the Khoisan languages) have rich systems of click consonants. Some geneticists have argued that certain aspects of the anatomy of speakers of these languages was selected for [30], and, aside from one language, Damin (potentially a ritual conlang), spoken by the men of the Lardil tribe in Northern Australia, there are no languages outside sub-Saharan Africa that use click consonants as an integrated part of their phonological system ([31], see also [32]).

Incorporating these two ideas into a simple, plausibly minimal, system of consonant articulation, with maximally distant places of articulation in the oral cavity, I developed the sound system for the language (table 4).

Table 4.

Tan!aa Kawawa ki consonants.

	bilabial	alveolar	palatal	velar
nasals	m	n	ny /ɲ/	ng /ŋ/
prenasalized clicks	m! /mʘ/	n! /nǀ/	–	–
oral stops	p	t	c	k
prenasalized stops	mb /mb/	nd /nd/	nj /ɲɟ/	ngg /ŋg/

I provided the production team with a simple way of writing the language, using combinations of letters for sounds (e.g. the palatal nasal was written as ny, the velar nasal as ng, and either click was !, with the pronunciation varying depending on the prenasalization place of articulation). I also provided the IPA symbols (e.g. ɲ, ŋ, ʘ, etc.), as well as recordings of the relevant examples. Following the direction of the discussion about nasality and clicks above, I decided to have only two clicks, both prenasalized, and, rather than a voicing distinction (between say /t/ and /d/), I adopted prenasalization for stops as the distinctive difference (between /t/ and /ⁿd/, where prenasalization of /t/ leads to obligatory voicing of /t/ to [d] under an assimilatory process (with potential future development to a full voicing contrast).

The minimal vowel system in the world's languages consists of three maximally acoustically distinct vowels, which were chosen as a maximally simple default [33], and semi-vowel versions of i and u (y [j] and w [w]) were added to the system. Vowel length allows extra vowel distinctions without increasing the places of articulation, so I adopted a two-way vowel length system, represented by simply doubling the vowel letter (table 5).

Table 5.

Tan!aa Kawawa ki vowels.

	front	centre	back
high	i [i] ii [i:]		u [u] uu[u:]
mid
low		a [a] aa[a:]

With the idea in mind that the language would be young, I gave it the simplest possible syllable structure, the structure of those syllables that babies make early in their utterances: either a simple vowel or a single consonant followed by a vowel ((C)V). These choices give the language, Tan!aa Kawawa ki (literally Mouth at People) a particular phonological organization which contributes to the overall aesthetics of the sound of the language. Initial syllables of roots are psychologically prominent in both processing and acquisition [34], and often attract stress in the world's languages [35], which motivated the decision to have these as the most prominent stresses in the language.

(b). Morphology and syntax

The morphosyntax and phrasal syntax of the language also followed these same basic design ideas: maximally simple with modern expressive capacity. For example, the pronominal system is the minimal system we see in languages across the world: it distinguishes between first person (ma) and second (mba) (as do, for example, Kiowa and Salt Yui, [36]). There is no specific third person pronoun (he/she/it), rather a demonstrative (ku), meaning something like ‘that (one)’ is used. Part of the motivation for this is the importance of pointing in human cognition: demonstratives are essentially acts of verbal pointing, hence likely to be an early innovation in language [37] and experimental work has shown them to be closely connected to pointing [38]. Demonstratives are also learned very early by children [39] and many languages today use demonstratives instead of third person pronouns (e.g. Kiowa, as mentioned above). Combining these ideas from cognitive science and linguistic typology allowed me to keep the pronominal system of the language as minimal as possible, while still providing the resources for rich communication.

In terms of word order, the conlang team had decided to use an idea from experimental work which shows that humans, irrespective of the order of words in their native language, prefer to provide the participants of a non-verbally presented action before specifying the action [40], an idea prefigured by Newmeyer's [41] claim that ‘Proto-World’ had a subject-object-verb order (also [21]). Emergent sign languages (where deaf children have to effectively create the language of a newly developed community) predominantly have subject-object-verb order too (e.g. [42]). This evidence for a cognitive bias towards that order led the team to adopt subject-object-verb as the order of the language.

I also made other choices about syntax as more complexity was required by the docudrama production team. Generally, the embedding of one clause within another is avoided in oral discourse [43], and many languages do not allow the embedding of one clause inside another [44], so the syntax of Tan!aa Kawawa ki uses periphrasis to express certain complex ideas. For example, I translated one of the lines from the script, ‘The only prey is where the tree-people are’, as follows:

(3)	Njima,	kawawa nunii,	wama ku ki.	Wu njima wama nyunga ki
	Food,	people tree,	place they loc.	Not food place else loc.

In English, the sentence has a fairly complex syntax, involving a focus-particle ‘only’, and a nominalized relative clause (‘where the tree-people are’) acting as part of a copular construction. In the language Tan!aa Kawawa ki, this comes out as something like ‘Food, the tree-people, at that/their place. No food at another place’, which avoids some of the complex syntax by juxtaposing two different clauses, and making the negative contrastive semantics of ‘only’ explicit. It also avoids another kind of embedding: rather than explicitly expressing ‘at the tree-people's place’, with an embedded possessive, it states both ‘food’ and ‘tree-people’ as topics, marked off by intonational boundaries, and the latter is then picked up by the demonstrative ku, meaning ‘that’ but referring to the tree-people.

Another example of this same idea appears in conditional clauses. One of the lines of the script was ‘If the tree-people are why we are hungry, deal with them!’. Semantically, this involves setting up a situation ‘the tree people cause our hunger’, and then stating the consequences of this (deal with them). Again, avoiding complex embedding, I instead used a pragmatically rich grammatical sentence-final particle [45], which is used to license the inference from a bad situation to its consequences. The translation is then:

(4)	Kawawa	nunii	unjima ndaa	ma ki,	ndiwu.	Ku nguu, mba!
	People	tree	hunger make	we loc	ADVERSE.	Them kill, you!

I have translated the pronoun ma here as ‘us’, but there is no plural expressed on this pronoun itself. This same sentence could equally be used to express that the tree-people are causing me to be hungry. With that in mind, this example could be more literally rendered as something like ‘(I consider) the tree people making hunger at us (to be a bad situation). (So) kill them, you!’ The second (imperative) clause is licensed pragmatically as the consequence of the speaker's attitude to the stated situation marked by the adversative particle ndiwu. I developed a rich set of these pragmatically complex final particles and they are used in the language for many clause chaining effects that might otherwise be expressed by clausal embedding.

One final linguistic technique I used, motivated by a well-established linguistic principle that organizes how the meanings of words are related to each other in human languages, is metaphor and metonymy [46]. Here, for example, is how to say, in Tan!aa kawawa ki, ‘We have hunted the food/prey for two days':

(5)	Ma	njiima	mambana	n!aana	ni.
	We	food	we-hunt	sun	hand

After the verb, there are two nouns, the word for ‘sun’ and the word for ‘hand’. The latter signifies that we are talking about two of something (as we have two hands). Although it is the case that familiar languages like English have quite arbitrary words for numerals, many languages use contentful words to count [47]. The word for ‘sun’ is used metonymically to mean ‘day’, on the assumption that early human languages would use processes of metonymy and metaphor to create rich semantic webs.

4. Comparing Beama and Tan!aa Kawawa ki

From the descriptions of the two languages above, it should be evident that there are similarities, but also differences in their construction processes. Some of these are the result of differences in the research that Schreyer and Adger each incorporated, as well as the artistic expectations of each production.

Adger's language Tan!aa Kawawa ki was intended to be set between 100 000 and 60 000 years ago, in Africa, while Schreyer's language Beama was intended to be set 20 000 years ago in Europe (likely in France, Spain and Portugal). This difference impacted our processes of language creation; Adger relied on the ‘youth’ of his language to make decisions, plus research from cognitive science on plausible linguistic defaults in the human mind, while Schreyer based her work on proposed proto-languages for this time period, as well as input from the director of Alpha. Adger's language might be considered an a priori language while Schreyer's is an a posteriori language since it drew explicitly from the hypothesized proto-language of Nostratic. However, despite this chronological difference, and difference in creation manner, there are also similarities between the languages.

Beama has 32 consonants and 6 vowels, while Tan!aa Kawawa ki has 14 consonants and 3 vowels (two semi-vowels), although this becomes six vowels through lengthening. Beama includes a fuller range of both manner and place of articulation, but the languages both incorporate the most typologically frequent and unmarked places of articulation (bilabial, alveodental and velar) as well as the two unmarked semi-vowels [w and j] [48]. In terms of vowels, Beama and Tan!aa Kawawa ki both include the minimal vowel system [i, u and a], although Beama has additional places of articulation [e, æ and o] while Tan!aa Kawawa ki creates more vowel variation through lengthening. The languages also both include simple CV syllable structures, but for different reasons. Adger used research on the phonology of young infants, while Schreyer based her decision on the creative input of the film's director. One difference is the inclusion of ejectives versus clicks in the two languages, supported by two different types of academic research [20,30]. The overall sound systems of both conlangs incorporate a set of empirical hypotheses about what the phonology of an early human language may have been like.

In terms of morphology and syntax, both Schreyer and Adger used subject–object–verb basic word orders for similar reasons. Adger's pronoun system was maximally simple and only included first person (ma) and a second person (mba), both neutral in terms of grammatical number, and his system relied on demonstratives for third persons, appealing to a cognitive primacy of these elements. Schreyer's pronoun system was more complex, and included first, second and third person singular and plural pronouns, with complexity in inclusivity and animacy (table 3).

Gobbo [49] compared ‘planned’ languages (conlangs) with natural languages and argued that ‘planned languages are more regular than natural languages in their morphology, and that this feature can be considered a simplification of the system. In other terms, natural languages are more complex than planned ones on the morphological level’ [49, p. 50]. The languages we have developed are also morphologically regular (as can be seen from the discussion above) and, while this is partly a side effect of their development, it has been suggested in research on early languages that they may have been simpler than languages of today [50,51]. However, others have argued that modern languages spoken in esoteric contexts (with smaller speaker numbers, with smaller geographical areas, and with less linguistic contact with other groups) have more grammatical complexity than modern exoteric languages (with larger speaker numbers, with larger geographical spread and with linguistic contact with other groups) [52,53]. Wray & Grace [52, p. 543] argue that the language used in esoteric communities may be ‘the natural default setting for human language’ partly because the communicative situations of these languages led to them being, to a large extent, non-compositional. Conlangs built with this research in mind would have quite different properties from the ones we have presented here. One relevant case would be morphological complexity in sign languages, many of which are fairly young languages, but which have esoteric characteristics. Aronoff et al. [54] argue that such languages can be at the low end of certain kinds of morphological complexity, though Schembri et al. [55] point out that this may be a side effect of the sociolinguistic characteristics of sign languages.

Connected to this, both Adger and Schreyer were given dialogue of rich grammatical complexity by the writers of the productions they were engaged on, and had to produce solutions to how to handle these, given the basic design choices that had been motivated by cognitive, linguistic and historical research. Since both languages would be used only in spoken form, and spoken language typically lacks rich embedding, Adger and Schreyer both had to negotiate how to make their language design expressive in the appropriate way. Independently, they both used grammatical particles at clause edges, though it is notable that Beama's syntax is considerably more complex than that of Tan!aa Kawawa ki in how it negotiates embedding.

The comparison of these two independently developed conlangs, both of which were created drawing on research-guided hypotheses about particular properties of language and cognition, on linguistic theory, and on empirical work in historical and descriptive linguistics, shows that it is possible to gain a novel perspective on how prehistoric languages may have worked as holistic systems.

5. Conclusion

In her 2015 presidential address to the American Anthropological Association, Monica Heller described the potential for invented languages to be windows into alternative realties. She writes:

[t]he story of invented languages helps us grasp the shape-shifting dimension of imagining alternative worlds that is, I want to argue, the specialty of anthropology. The same two ideas have been used to try to solve all kinds of issues: either we think we will solve human problems by finding a way to communicate clearly across difference, or we think we will do so by embracing the incommensurability of difference. [56, p. 13]

While Heller was specifically focusing on Esperanto as her example, an auxlang, in this paper, we have examined if the study of constructed languages might address whether artlangs also have this capacity to provide alternative windows into prehistoric society. Attempting to understand how prehistoric languages may have worked does indeed mean we have to think about linguistic systems that may well have been incommensurably different from ours. However, as Heller notes, imagination can be a way to open up new perspectives on how different, or similar, the languages of such remote ancestors were. When we combine this imaginative perspective with research from cognitive science, historical linguistics, linguistic typology and comparative anthropology, we can paint a picture of how these languages may have worked as wholes, in terms of both their structural properties, and how they were used by their communities. We can also paint a picture of the world portrayed in the media production and provide a sense of authenticity. We are not claiming any kind of verisimilitude, of course, but rather arguing that conlanging can be an interesting way to open our eyes to a larger vista beyond the fragmentary knowledge that we can glean from research about this most intractable area of language.

Data accessibility

This article has no additional data.

Authors' contributions

C.S. and D.A. contributions were equal to design, analysis and interpretation of data, drafting and revising the article and final approval for submission.

Competing interests

We declare we have no competing interests.

Funding

We received no funding for this study.